Horticulture Research, 2022, 9: uhac042 https://doi.org/10.1093/hortre/uhac042 Article New insights into f lowering date in Prunus: fine mapping of a major QTL in sweet cherry Camille Branchereau1 , José Quero-García1 , *, Nathalia Helena Zaracho-Echagüe2 , 3 , Laurine Lambelin1 , Mathieu Fouché1 , Bénédicte Wenden1 , Armel Donkpegan4 , Loïck Le Dantec1 , Teresa Barreneche1 , David Alletru5 , Julien Parmentier5 and Elisabeth Dirlewanger1 , * 1 INRAE, Univ. Bordeaux, UMR Biologie du Fruit et Pathologie, 33882 Villenave d’Ornon, France 2 Centre for Research in Agricultural Genomics (CRAG) CSIC-IRTA-UAB-UB, Campus UAB, 08193 Bellaterra, Barcelona, Spain 3 IRTA, Centre de Recerca en Agrigenómica CSIC-IRTAUAB-UB, Campus UAB, Bellaterra, 08193 Barcelona, Spain 4 SYSAAF-Centre INRAE Val de Loire, UMR BOA, 37380 Nouzilly France. 5 INRAE, UE 0393, Unité Expérimentale Arboricole, F-33210 Toulenne, France * Corresponding author: E-mail: elisabeth.dirlewanger@inrae.fr, jose.quero-garcia@inrae.fr Abstract Flowering date is an important trait in Prunus fruit species, especially for their adaptation in a global warming context. Numerous quantitative trait loci (QTLs) have been identified and a major one was previously located on LG4. The objectives of this study were to fine-map this QTL in sweet cherry, to identify robust candidate genes by using the new sweet cherry genome sequence of the cultivar “Regina” and to define markers usable in marker-assisted selection (MAS). We performed QTL analyses on two populations derived from crosses using cultivars “Regina” and “Garnet” as parents. The first one (n = 117) was phenotyped over ten years, while the second one (n = 1386) was evaluated during three years. Kompetitive allele specific PCR (KASP) markers located within the QTL region on LG4 were developed and mapped within this region, consisting in the first fine mapping in sweet cherry. The QTL interval was narrowed from 380 kb to 68 kb and candidate genes were identified by using the genome sequence of “Regina”. Their expression was analyzed from bud dormancy period to f lowering in cultivars “Regina” and “Garnet”. Several genes, such as PavBOI-E3, PavSR45a and PavSAUR71, were differentially expressed in these two cultivars and could be then considered as promising candidate genes. Two KASP markers were validated using a population derived from a cross between cultivars “Regina” and “Lapins” and two collections, including landraces and modern cultivars. Thanks to the high synteny within the Prunus genus, these results give new insights into the control of f lowering date in Prunus species and pave the way for the development of molecular breeding strategies. Introduction Flowering in perennial plants is dependent on bud dormancy, an important evolution strategy to face and In temperate fruit tree species, f lowering date (FD) is a survive under unfavorable climatic conditions and that trait of main importance and highly dependent on the cli- allows plants to grow and bloom under optimal condi- mate conditions of the production area. In sweet cherry tions [5]. During fall and winter, when daylight and tem- (Prunus avium L.), breeding strategies for crop adaptation peratures decrease, trees enter endodormancy: internal/- aim at the development of early or late blooming cul- physiological factors prevent growth even under optimal tivars [1]. On the one hand, late blooming cultivars are conditions. Buds exit this deep stage of dormancy only requested in cold regions in order to avoid frost damages after a certain amount of low temperatures is accumu- in early spring. On the other hand, early blooming ones lated by the plants (fulfillment of chilling requirements, are preferred in warmer regions in order to avoid high CRs). Following the endodormancy release, in late winter temperatures during the f lowering period, which could and beginning of spring, trees enter ecodormancy, which decrease the fertility of cultivated plants by reducing implies control by external/environmental factors, such stigmatic receptivity, pollen germination and pollen tube as temperature and photoperiod. Bud development is growth and therefore induce low fruit set [2,3]. Moreover, prevented until optimal conditions are met later in spring as FD is at least partially correlated to maturity date [4] and trees accumulate a sufficient amount of warm tem- (harvest time), breeders generally look for a large range of peratures (fulfillment of heat requirements, HRs) to over- FD, so that fruit ripening spreads over time. In particular, come dormancy. Flowering can finally occur. CRs and breeders seek extra-early ripening cultivars, for which HRs estimations are time-consuming since many twigs fruit can reach very high prices [1]. need to be sampled and specialized equipment such as Received: 30 July 2021; Accepted: 4 February 2022; Published: 19 February 2022; Corrected and Typeset: 1 May 2022 © The Author(s) 2022. Published by Oxford University Press on behalf of Nanjing Agricultural University. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (https://creativecommons.org/licenses/by/4.0/), which permits unrestricted reuse, distribution, and reproduction in any medium, provided the original work is properly cited. 2 | Horticulture Research, 2022, 9: uhac042 climatic chambers is required for experiments in forcing growth to enter dormancy despite dormancy-inducing conditions. For these reasons, FD is often the single trait conditions [18,19]. In sweet cherry, DAM 5 and 6 were analyzed. This is particularly the case when large num- considered to be strong candidate genes for the QTLs for bers of individuals have to be studied. FD and CRs on LG1 [20]. For the QTL on LG4, the most The phenological cycle of tree species is synchronized promising candidate genes were related to gibberellin with alternating seasons and environmental conditions and temperature sensing pathways [20]. These candidate and therefore the succession of low and warm tem- genes were identified using the peach genome sequence peratures in winter and spring, respectively, is essential annotation [21]. for flowering. Disruptions due to climate warming have The aims of this study were i) to detect stable QTLs already been noticed in temperate tree species like sweet across numerous years for FD, ii) to fine-map the major cherry [6]. Indeed, reductions in available winter chill QTL on LG4, iii) to identify candidate genes within the and the non-satisfaction of CRs can induce low fruit set, reduced QTL interval and characterize their expression, putting fruit production at risk and leading to important and iiii) to develop markers usable for breeding selection. economic losses [6,7]. Investigating the genetic determin- This work should contribute to increase the efficiency ism of FD is therefore highly relevant in order to maintain of breeding programs for sweet cherry and other Prunus the production in temperate climates. species to create new cultivars well adapted to the future To date, numerous quantitative trait locus (QTL) analy- climatic conditions. ses on sweet cherry and other species belonging to Prunus genus have led to a better understanding of the genetics of this trait [4,8–10]. Moreover, the high genomic synteny Results of Prunus species often leads to the detection of QTLs Flowering date evaluation in similar chromosomal regions [11]. It is well known Populations #1 and #2 (Table 1), as well as the parental now that FD is a quantitative trait with high broad sense cultivars “Regina” and “Garnet”, were evaluated for three heritability [4,8]. Moreover, several studies have showed FD stages, beginning of flowering (BF), full flowering (FF) that FD seems to be more dependent on CRs than on HRs and end of flowering (EF), across several seasons from in sweet cherry [8,12], as well as in other Prunus species 2008 to 2021 (Fig. 1 for FF, Fig. S1 for BF and EF and like almond [10,13], apricot [14] and peach [15]. FD and Table S1) characterized by contrasted temperatures dur- CRs are highly correlated in Prunus species [7]. ing the endodormancy and ecodormancy periods (Fig. S2 Although QTLs for FD were detected on all linkage and Table S2). For instance, during the month of Decem- groups (LGs) in sweet cherry, major loci were located on ber (endodormancy), the mean temperatures varied from LGs 1 and 4 [4,8,9]. Using two sweet cherry F1 populations 4.6◦ C (season 2010–2011) to 9.1◦ C (season 2011–2012) from crosses “Regina” × “Garnet” and “Regina” × “Lapins”, whereas during the month of February (ecodormancy), Castède et al. [8] detected major QTLs for both CRs and the mean temperatures varied from 1.7◦ C (season 2011– FD, stable over the years of evaluation, and in the same 2012) to 11.6◦ C (season 2020–2021) (Table S2). In 2021, region of LG4. Such colocalizations were also found on EF could not be scored because of frost events in early LGs 1 and 7 of parents “Lapins” and “Regina”, respectively, spring. In Population #2, FDs of reciprocal crosses (R × G and confirmed the high correlation between both traits and G × R) were compared and no significant statistical and the importance of CRs for FD. A minor QTL for HRs differences were found (data not shown). Therefore, all was also found within the LG4 region. Cai et al. [16] hybrids were grouped and analyzed together in the fol- identified QTLs for FD in three F1 populations of sour lowing sections. cherry (Prunus cerasus), among them two were located on FD was highly dependent on the year of evaluation LGs 1 and 4. This detection on sour cherry is relevant (Fig. 1). In Population #1, years of evaluation could be for QTL analyses in sweet cherry since half of the sour classified into distinct groups with significant differences cherry genome is derived from sweet cherry, the sour (results of the statistical tests not shown) of FD: 2017, cherry genome being divided in two subgenomes from 2011, 2008–2009-2014, 2012, 2010, 2013–2015-2016, from two Prunus species, “avium” subgenome and “fruticosa” early to late FD (Fig. 1A). In Population #2, FD was sig- subgenome [17]. Calle et al. [9] used six sweet cherry nificantly different across the three years of evaluation populations (four F1 and two F2 ), most of them obtained (Fig. 1). The average monthly temperatures across the from the early blooming cultivar “Cristobalina”, with low seven years with early FD (2008, 2009, 2011, 2014, 2017, CRs [12], and detected a major QTL for FD on LG1. This 2019 and 2021) were: 13.3◦ C in October, 8.9◦ C in Novem- locus overlapped with QTLs for CRs and FD described in ber, 6.1◦ C in December, 5.7◦ C in January and 8.7◦ C in Castède et al. [8]. February. Across the six years with late FD (2010, 2012, Candidate genes have been suggested for QTLs on 2013, 2015, 2016 and 2018), the average monthly tem- LGs 1 and 4 [18–20]. The QTL region on LG1 carries the peratures were: 14.4◦ C in October, 10.9◦ C in November, DORMANCY-ASSOCIATED MADS-box (DAM) genes. These 7.5◦ C in December, 6.5◦ C in January and 5.2◦ C in Febru- genes, six in number (DAM1–6), were initially studied in ary. Hence, flowering occurred earlier when tempera- the non-dormant evergrowing peach mutant (evg), which tures were lower from October to January and higher in presents a deletion in this region and does not cease February. Branchereau et al. | 3 Table 1. Characteristics of the two F1 populations used in this study for the QTL analyses Population #1 Population #2 General information Cross “Regina” × “Garnet” “Regina” × “Garnet” (R × G) and “Garnet” × “Regina” (G × R) Crossing method Manual pollination Bumblebees in confined tunnels Year of the cross 2001 2010 Year of plantation (on own roots) 2003 2012 Number of individuals 117 1386 (793 R × G and 593 G × R) FD phenotyping Number of years 10 years: 2008–2012 [8] and 2013–2017 3 years: 2018, 2019 and 2021 new phenotyping Flowering stages scored Beginning, Full and End (BF, FF and EF) Beginning, Full and End (BF, FF and EF) Number of phenotyped individuals 117 1386 Genotyping SNP array size (RosBREED SNP chips) 6 K [8,22] 6 + 9 K [23] Number of genotyped individuals 117 454 Figure 1. Distribution of f lowering date in Population #1 (from 2008 to 2017) and Population #2 (in 2018, 2019 and 2021). A, box plot of full f lowering distribution scored in Julian days (JDs) across ten years in Population #1 (in grey) and three years in Population #2 (2018 in green, 2019 in orange and 2021 in purple). B, distribution of full f lowering scored in JDs across three years in Population #2 (2018 in green, 2019 in orange and 2021 in purple). Data for parental cultivars “Regina” (R) and “Garnet” (G) in 2018, 2019 and 2021 is indicated by arrows in B. Every season, “Garnet” was the first parent to bloom. We calculated broad-sense heritability (H2 ) for each For instance, in 2018 and 2019, FF for “Garnet” was respec- flowering stage in both populations. In Population #1, H2 tively seven and thirteen days earlier than “Regina” (“Gar- was equal to 0.96 for BF, FF and EF. In Population #2, net”: 99 and 88 Julian Days (JDs, i.e. number of days from heritabilities were lower: 0.90 for BF, 0.88 for FF (both January 1st ), “Regina”: 106 and 101 JDs) (Fig. 1B). calculated using three years of measurements) and 0.77 Flowering stages (BF, FF and EF) were highly correlated for EF (calculated with two years of measurements). in both populations (from 0.59 to 0.97 in Population #1 and from 0.66 to 0.94 in Population #2) (Table S3). Cor- Linkage maps relations were higher between different f lowering stages New genetic maps of “Regina” and “Garnet” were within a single year (for instance in Population #2, from constructed using a subset of Population #2 (454 0.75 to 0.80 in 2018, from 0.80 to 0.90 in 2019, and 0.94 R × G hybrids). After filtering SNPs according to their in 2021) than for a single stage across years (from 0.76 to quality and low missing data, 1619 SNPs were retained. 0.78 for BF, from 0.68 to 0.71 for FF, and 0.71 EF). Among these, 598 SNPs were heterozygous in “Regina” 4 | Horticulture Research, 2022, 9: uhac042 (), 446 heterozygous in “Garnet” (), and #1 (2 126 110 bp) and less than 0.5 cM (378 518 bp) in 575 were heterozygous in both parents (). As we Population #2 (Table 2). constructed parental linkage maps and not a consensus map of both parents, we did not use the heterozygous Fine mapping of the QTL on LG4 of “Regina” markers in both parents. Genetic maps of each parent The fine mapping of the “Regina” LG4 region was carried are described in Table S5 and Fig. S3. The map obtained out in two steps. Firstly, parental cultivars and the for “Regina” included a higher number of markers than whole Population #2 (1386 individuals) were genotyped the one for “Garnet” (598 and 446 markers respectively). with nine KASP markers: KASP_9.269, KASP_9.271, However, genetic lengths of both maps were similar. The KASP_9.456, KASP_9.727, KASP_9.780, KASP_9.890, KASP_ two largest LGs were the LGs 1 of both parents, called 9.935, KASP_10.090 and KASP_10.159 (Tables 3 and R1 for “Regina” and G1 for “Garnet”, with 78 markers S4). These nine KASP markers were integrated in the covering 136 cM and 183 markers covering 167.4 cM, genetic map of the LG4 of “Regina” (Fig. 2). “Garnet” was respectively. The average distance between markers in homozygous whereas “Regina” was heterozygous for “Regina” and “Garnet” parental maps were equal to 1.1 the nine KASP markers. Among the 1379 individuals for and 1.7 cM, respectively. Several large gaps were also which genotypes were obtained for all these KASPs, 1338 found in the maps, especially on LGs R6 (gap = 35.5 cM), (97%) were non-recombinant: 641 were homozygous for G2 (32.3 cM) and G4 (31.1 cM) (Table S5). the nine KASP markers (as observed in “Garnet”) and 697 were heterozygous for the nine KASP markers (as QTL analyses for flowering date observed in “Regina”). The other 41 individuals were QTL analyses were performed for BF, FF and EF in recombinant (i.e. with one recombination event between both populations, with year-by-year and multi-year two markers) (Tables 3 and S4). The second step was approaches. Due to the high correlation between the to genotype these 41 recombinant individuals with three f lowering stages, only the QTLs for FF are pre- eight new KASPs: KASP_9.781, KASP_9.801, KASP_9.814, sented, as in Castède et al. [8], and are thereafter called KASP_9.916, KASP_9.933, KASP_9.936, KASP_9.958 and “qP-FD”. KASP_9.970 (Tables 3 and S4) to further increase fine Concerning Population #1, QTLs were found on all mapping accuracy. Seventeen KASP markers were LGs of “Regina” and on LGs G1, G2, G3, G5, G6 and G8 therefore used for fine mapping. of “Garnet” using year-by-year and multi-year analyses Recombinant individuals were grouped into 12 dis- (Tables 2 and S6). Thirteen QTLs were detected with the tinct recombinant genotypes (called Rec #1 to Rec multi-year analysis based on ten years of data (Table 2). #12) depending on the position of the recombination Only the loci on LGs R4 (qP-FD4.1m ), R7 (qP-FD7.1m ), events (Tables 3 and S8). No recombination event G1 (qP-FD1.2m ) and G6 (qP-FD6.2m ) explained more occurred between several markers: KASP_9.269 and than 5% of the phenotypic variation in the multi-year KASP_9.271; KASP_9.727 and KASP_9.781; KASP_9.801 analysis. and KASP_9.814; KASP_9.916 and KASP_9.933; KASP_9.935 With Population #2, QTLs were detected on almost and KASP_9.936; KASP_9.958 and KASP_9.970; and all LGs of both parents: all LGs of “Regina” (R1 to R8) KASP_10.090 and KASP_10.159 (Table 3). Genotypes at and LGs G1, G2, G3, G7 and G8 of “Garnet” (Tables 2 the 17 KASPs and least square means (lsmeans) for FD and S7). Eleven QTLs were significant in the multi- for the parents, the non-recombinant genotypes and the year analysis and most of them were stable across twelve recombinant genotypes are presented in Table 3. years in single-year analyses (shaded in grey in Tables 2 FD values in 2018, 2019 and 2021 of the individuals of and S7). In the multi-year analysis, only the QTLs Population #2 are presented in Table S8. located on LGs R4 (qP-FD4.1m ) and G1 (qP-FD1.2m ) Non-recombinant individuals homozygous for all the explained more than 5% of the phenotypic varia- 17 KASP markers, as “Garnet”, presented a FD similar tion (Table 2). QTLs on LGs G5 and G6 (qP-FD5.2m , to “Garnet” itself, with FF lsmeans equal to 90.5 and qP-FD6.2m ) were only detected with Population #1 90.7 JDs, respectively (Table 3). In comparison, the dif- and QTL on LG G7 (qP-FD7.2m ) was only found with ference between FD of the non-recombinant individuals Population #2. heterozygous for the 17 KASP markers, as “Regina”, and In both populations, the QTL on LG R4, qP-FD4.1m , “Regina” itself, was much more important: in average, explained the largest PVE (Fig. 2 and Table 2). In Pop- seven days of difference were found (FF lsmeans equal to ulation #1, it was the only locus to be significant in 93.6 and 100.7 JDs for hybrids and “Regina”, respectively) every single-year analysis, across the 10 years (Fig. 2 (Table 3). “Regina” was flowering much later than the and shaded in grey in Tables 2 and S6). With the multi- hybrids with the same non-recombinant genotype. The year analysis, PVE were equal to 34.3% and 37.1% in difference between the FDs of non-recombinant individ- Populations #1 and #2, respectively (Table 2). It reached uals as “Garnet” and of the non-recombinant individuals 46.3% in 2015 in Population #1 and 39.8% in 2019 in as “Regina” was rather small, only three days on average, Population #2 (Tables S6 and S7). The QTL qP-FD4.1m was but statistically significant (90.5 and 93.6 JDs, Table 3). detected within smaller confidence intervals in Popu- Among the 41 recombinant individuals, eight pre- lation #2 than in Population #1: 3.2 cM in Population sented intermediate phenotypes that could not be Branchereau et al. | 5 Table 2. Flowering date quantitative trait loci (QTLs) detected with multi-year analyses in Populations #1 (across ten years) and #2 (across three years) QTL name LG L (cM) CI 95% (cM) Physical LOD PVE mean d mean Nb of years position (Mb) (%) where significant Population #1: 117 hybrids - 10 years of evaluation (2008–2017) qP-FD1.1m R1 12.8 0–32.7 0.48–11.89 18.9 4.0 −0.9 1 qP-FD2.1m R2 29.8 27.2–32.3 26.33–27.77 19.5 3.7 −0.9 0 qP-FD4.1m R4 20.6 19–22.2 8.99–11.12 146.1 34.3 2.9 10 qP-FD5.1m R5 19.0 0–42.4 3.30–14.88 11.4 2.1 0.6 0 qP-FD6.1m R6 66.4 36.4–81.7 10.96–31.64 18.7 3.4 −0.8 3 qP-FD7.1m R7 54.6 34.7–57.6 22.16–28.17 31.9 5.7 1.1 7 qP-FD8.1m R8 18.1 0–43.3 1.91–18.94 8.8 1.6 −0.4 0 qP-FD1.2m G1 128.7 101.7–151.2 39.91–54.07 24.1 7.3 1.3 4 qP-FD2.2m G2 15.3 3.8–26.7 2.59–13.18 17.2 4.7 −1.0 0 qP-FD3.1m G3 88.8 32–100.3 6.96–29.84 12.2 3.4 0.7 0 qP-FD5.2m G5 11.3 0–25.6 6.82–9.10 11.4 3.1 −0.8 0 qP-FD6.2m G6 20.7 16.9–24.4 3.07–5.79 38.5 12.8 −1.9 2 qP-FD8.2m G8 58.0 29.1–72.1 12.40–22.82 12.2 3.1 0.9 0 Population #2: 454 hybrids - 3 years of evaluation (2018–2019-2021) qP-FD1.1m R1 26.9 21.2–32.6 9.85–11.96 21.1 3.3 1.1 3 qP-FD2.1m R2 23.1 15.8–30.5 23.72–27.82 12.3 1.9 0.8 3 qP-FD4.1m R4 26.9 < 0.5 cM 9.78–10.16 180.3 37.1 −3.6 3 qP-FD5.1m R5 35.9 20.1–51.7 10.27–18.77 17.8 2.8 1.0 3 qP-FD6.1m R6 55.1 25.2–84.9 9.59–30.27 16.2 3.5 −1.1 3 qP-FD7.1m R7 53.8 36.2–61.6 23.74–29.42 19.7 3 −1.0 3 qP-FD8.3m R8 60.7 53.9–67.4 20.01–25.42 14.1 2.2 −0.8 2 qP-FD1.2m G1 155.5 150.1–160.9 49.13–50.88 20.0 6.1 −1.5 3 qP-FD2.2m G2 33.4 0.0–84.2 0.53–33.79 9.8 2.9 −0.7 2 qP-FD7.2m G7 8.0 0.0–17.2 0.36–12.88 7.6 2.2 0.8 2 qP-FD8.2m G8 26.7 8.3–45.1 2.45–17.36 6.0 1.7 0.7 1 LG, linkage group; L, distance from the beginning of the chromosome to the point of maximum LOD in the interval; CI, confidence interval; Physical position of flanking markers on “Regina” v1 genome sequence in mega base pairs (Mb); LOD, logarithm of the odds ratio; PVE mean, mean value of PVE (phenotypic variance explained by the QTL in percentage of the total variation) in the multi-environment analysis; d mean, mean value of d (difference X(A) – X(B) according to the year of evaluation, where A and B are the two homozygotes at the marker loci) in the multi-environment analysis; (+/), the sign varies according to the year of evaluation); Nb of years where significant, number of years where the QTL was detected in single-year analysis. QTLs detected every year are shaded in grey. clearly assigned to early or late f lowering classes (six individuals) were statistically different (p < 0.05, and consequently could not be used to fine-map Table 3). Based on 30 individuals, these results indicated the QTL (Table 3). These individuals belonged to the that the QTL was located downstream KASP_9.890. A recombination groups Rec #1 (two individuals), #2 (one), single individual with Rec #10 genotype indicated that #5 (one) and #7 (four). Therefore, only the remaining the QTL was located downstream KASP_9.935. On the 33 individuals recombining between KASP_9.456 and other side, one individual with Rec#11 genotype indicated KASP_10.090 (Rec #3, #4, #6, #8, #9, #10, #11 and #12, that the QTL was located upstream (before) KASP_9.958. Table 3) were used for the fine mapping. Taken all together, these results indicate that the QTL Seventeen individuals with Rec #3 and Rec #4 geno- is located between KASP_9.890 and KASP_9.958, within types indicated that the QTL was located downstream a region of 68 kb (Table 3). FD of individuals that were (after) KASP_9.456: individuals that were homozygous homozygous in this interval (16 individuals with Rec#1, 3 from KASP_9.727 to KASP_10.159 (Rec #3) were early and 6) was significantly different from FD of individuals flowering while individuals that were heterozygous (Rec that were heterozygous (14 individuals with Rec#2, 4, 5, 7 #4) were late f lowering. FDs of these two recombinant and 12) (p-value = 0.0087). groups (Rec#3 with 10 individuals and Rec#4 with 7 individuals) were statistically different (p < 0.01, Identification of candidate genes for the QTL on Table 3). Thirteen individuals with Rec #6 genotype LG4 (early f lowering and homozygous from KASP_9.890 to According to the QTL analysis, the major QTL qP-FD4.1m KASP_10.159), Rec #8 genotype (early f lowering and on the LG R4 mapped between markers KASP_9.780 homozygous from KASP_9.916 to KASP_10.159) and Rec (9 780 346 bp) and KASP_10.159 (10 158 864 bp). We #9 genotype (late f lowering and heterozygous from identified 65 predicted genes within this region of KASP_9.916 to KASP_10.159) indicated that the causal less than 380 000 bp (Table S9). Using transcriptomic region was located downstream KASP_9.890. FDs of data obtained with RNA-sequencing for the cultivars recombinant groups Rec#8 (three individuals) and Rec#9 “Regina” and “Garnet” during bud dormancy over two 6 | Table 3. Genotypes at the 17 KASPs and phenotypes (least-square mean values for Full f lowering, FF, in Julian Days) of the individuals of Population #2 KASP KASP_ KASP_ KASP_ KASP_ KASP_ KASP_ KASP_ KASP_ KASP_ KASP_ KASP_ KASP_ KASP_ KASP_ KASP_ KASP_ KASP_ FD Nb of Statistical 9.269 9.271 9.456 9.727 9.780 9.781 9.801 9.814 9.890 9.916 9.933 9.935 9.936 9.958 9.970 10.090 10.159 ind. significance Position 9269 9271 9456 9727 9780 9781 9801 9814 9890 9916 9933 9935 9936 9958 9970 10 090 10 159 FF of the Horticulture Research, 2022, 9: uhac042 (kb) lsmeans difference G T/T A/A A/A A/A A/A C/C A/A G/G C/C G/G C/C G/G A/A C/C A/A A/A T/T 90.7 G - R C/T G/A A/T A/G A/G C/G A/G G/A A/C G/A C/T A/G A/G C/T A/T A/G T/C 100.7 R Non-rec. T/T A/A A/A A/A A/A C/C A/A G/G C/C G/G C/C G/G A/A C/C A/A A/A T/T 90.5 641 ∗∗∗ Like G (<2.2e -16) Non-rec. C/T G/A A/T A/G A/G C/G A/G G/A A/C G/A C/T A/G A/G C/T A/T A/G T/C 93.6 697 Like R Rec #1 C/T G/A A/A A/A A/A C/C A/A G/G C/C G/G C/C G/G A/A C/C A/A A/A T/T 91.8 2 - Rec #2 T/T A/A A/T A/G A/G C/G A/G G/A A/C G/A C/T A/G A/G C/T A/T A/G T/C 92 1 Rec #3 C/T G/A A/T A/A A/A C/C A/A G/G C/C G/G C/C G/G A/A C/C A/A A/A T/T 90.8 10 ∗∗ Rec #4 T/T A/A A/A A/G A/G C/G A/G G/A A/C G/A C/T A/G A/G C/T A/T A/G T/C 93.5 7 (0.009396) Rec #5 T/T A/A A/A A/A A/A C/C A/G G/A A/C G/A C/T A/G A/G C/T A/T A/G T/C 92 1 - Rec #6 C/T G/A A/T A/G A/G C/G A/G G/A C/C G/G C/C G/G A/A C/C A/A A/A T/T 90.7 4 NS Rec #7 T/T A/A A/A A/A A/A C/C A/A G/G A/C G/A C/T A/G A/G C/T A/T A/G T/C 91.5 4 Rec #8 C/T G/A A/T A/G A/G C/G A/G G/A A/C G/G C/C G/G A/A C/C A/A A/A T/T 91.3 3 ∗ (0.03725) Rec #9 T/T A/A A/A A/A A/A C/C A/A G/G C/C G/A C/T A/G A/G C/T A/T A/G T/C 94.1 6 Rec #10 C/T G/A A/T A/G A/G C/G A/G G/A A/C G/A C/T G/G A/A C/C A/A A/A T/T 91 1 Rec #11 C/T G/A A/T A/G A/G C/G A/G G/A A/C G/A C/T A/G A/G C/C A/A A/A T/T 92.7 1 - Rec #12 C/T G/A A/T A/G A/G C/G A/G G/A A/C G/A C/T A/G A/G C/T A/T A/A T/T 94.7 1 FF lsmeans values are presented for the parental cultivars ‘Garnet’ (G) and ‘Regina’ (R), the non-recombinant individuals (Non-rec. Like G and Non-rec. Like R) and the recombinant genotypes (Rec #1 to Rec #12). The number of individuals is presented in column ‘Nb of ind.”. The physical position of the seventeen KASP markers on the ‘Regina’ genome is given in line ‘Position (kb)’ in kilo base pairs (kb). ‘Garnet’ is homozygous for all markers, ‘Regina’ is heterozygous for all markers. ∗∗∗, p < 0.001; ∗∗, p < 0.01; ∗, p < 0.05; NS, p > 0.05; −, not enough individuals to perform reliable statistical test. The region surrounded in black indicates the fine-mapped region. Branchereau et al. | 7 Figure 2. Quantitative trait loci (QTLs) for f lowering date (FD) detected on “Regina” LG4 in both Populations #1 (left) and #2 (right). QTLs detected using the single-year analyses are named “FD_20xx”. QTLs detected using the multi-year analysis method are named “FD_MultiYear” and written in bold. Percentage of phenotypic variance explained (PVE) is given for each QTL. Homolog markers between both maps are linked by a line. In the map obtained with Population #2, on the right, the first set of nine KASP markers created to saturate the region of the QTL for FD and for fine mapping are colored in blue. seasons (2009/2010 and 2015/2016) [24], we found that and the small auxin-up RNA (SAUR) protein SAUR71 19 out of the 65 genes were not expressed in our (PAV04_ REGINAg0203401). We renamed them PavBOI-E3, plant material. The 46 remaining genes were expressed PavSR45a and PavSAUR71, respectively (Fig. 3). with different patterns throughout dormancy (Table S9, PavBOI-E3 was slightly more expressed in “Regina” than Fig. S4). in “Garnet” during endodormancy of both cultivars (prior Based on the fine mapping, the QTL region could be February). However, after “Garnet” dormancy release, redefined into a smaller region between KASP_9.890 expression of this gene significantly increased in this (9 889 761 bp) and KASP_9.958 (9 957 756 bp), covering cultivar. After the dormancy release of “Regina”, the around 68 kb. Twelve candidate genes were located gene remained more expressed in “Garnet”. Levels of within this new interval (shaded in grey in Table S9). expression of PavSR45a increased during dormancy and Using transcriptomic analyses, we looked for genes after “Garnet” dormancy release in both cultivars, and that were differentially expressed in “Regina” and decreased in both cultivars after “Regina” dormancy “Garnet” during endodormancy and ecodormancy, and release. The expression decline was more important that could explain the FD phenotypic differences in “Garnet”. PavSAUR71 was not expressed during we observed in these cultivars. Among the twelve endodormancy and its expression started to increase candidate genes, three exhibited differences between after dormancy release in both parental cultivars, the the two cultivars in their expression profiles after increase being more important in “Garnet”. For these endodormancy release. These candidate genes were three candidate genes, expression profiles obtained predicted to encode the BOI-related E3 ubiquitin-protein in both RNA-seq analyses were consistent (i.e. same ligase 3 (PAV04_REGINAg0203291), the serine/arginine- patterns between December and March, period in rich splicing factor SR45a (PAV04_ REGINAg0203371) common in both analyses). 8 | Horticulture Research, 2022, 9: uhac042 Figure 3. Expression profiles of three candidate genes of interest within the reduced interval of the f lowering date (FD) QTL on LG4: PavBOI-E3, PavSR45a and PavSAUR71. Profiles are from the 2015/2016 RNA-seq analysis [24] (A) and the 2009/2010 RNA-seq analysis (B). Expression levels were measured in transcripts per million (TPM) in parental cultivars “Regina” in blue and “Garnet” in red. In A, vertical dashed and solid lines correspond to the dormancy release dates and beginning of f lowering dates, respectively, for “Regina” (in blue) and “Garnet” (in red). Validation of two KASP markers both markers (55 hybrids A:A for KASP_9.936 and C:C for Four KASP markers (KASP_9.814, KASP_9.916, KASP_9.936 KASP_9.958) (Table 4). and KASP_9.958) were used to genotype an F1 population In both germplasm collection and set of cultivars, a derived from the cross between cultivars “Regina” and few individuals presented a new homozygous genotype “Lapins” (R × L, n = 115), accessions from the germplasm (G:G for KASP_9.936 and T:T for KASP_9.958) (Table 4). collection (n = 104) and a set of cultivars (n = 51) (Table S10). Due to the low number of these individuals, we did not Markers KASP_9.814 and KASP_9.916 presented a very take them into account for the statistical analyses. low level of polymorphism in the two sets of accessions. Concerning the germplasm collection, for KASP_9.936, In the germplasm collection, 92 accessions out of 104 43% of the individuals were heterozygous and 44% were were homozygous “G:G” (as “Garnet”) for KASP_9.814, and homozygous as “Garnet” (A:A) (Table 4). For KASP_9.958, 100 were homozygous “G:G” (as “Garnet”) for KASP_9.916. 47% of the individuals were heterozygous and 46% were In the set of 51 cultivars, 49 were homozygous “G:G” homozygous as “Garnet” (C:C) (Table 4). For both markers, for KASP_9.814 and all were homozygous “G:G” for heterozygous individuals were significantly later flow- KASP_9.814. Therefore, we could not use them to conduct ering than individuals with the “Garnet” homozygous statistical analyses. For this reason, only the results genotype (+ 1.7 days for KASP_9.936 and + 1.3 days for obtained with KASP markers KASP_9.936 and KASP_9.958 KASP_9.958) (Table 4). When considering the haplotypes are presented (Table 4). defined by the two KASP markers, heterozygous individ- In the R × L population, heterozygous individuals (52% uals for both markers were flowering 1.7 days later than of the population) were f lowering 2.9 and 3.0 days homozygous individuals for both markers. later than homozygous ones (48%) for KASP_9.936 and Concerning the panel of 51 cultivars, most individuals KASP_9.958, respectively (p-values <2.2e-16) (Table 4). presented the same genotype as “Regina” and “Garnet” When considering the haplotypes defined by the two as well. For KASP_9.936, 59% of the individuals were het- markers, heterozygous individuals for both markers (59 erozygous and 37% were homozygous as “Garnet” (A:A). hybrids G:A for KASP_9.936 and T:C for KASP_9.958) were For KASP_9.958, the heterozygous/homozygous distribu- flowering 2.9 days later than homozygous individuals for tion was 55% and 41% of the individuals (Table 4). Again, Branchereau et al. | 9 Table 4. Allelic frequency, phenotyping data and statistical analyses for two KASP markers in the “Regina” × “Lapins” population, a germplasm collection and a set of cultivars “Regina” × “Lapins” Germplasm collection Cultivars (n = 51) (n = 115) (n = 104) Nb ind. Average FD Nb ind. Average FD Nb ind. Average FD (BF lsmeans) (BF lsmeans) (BF lsmeans) KASP_9.936 Genotype A:A (as 55 92.0 45 91.1 19 87.0 “Garnet”) Genotype G:A (as 59 94.9 46 92.7 30 88.3 “Regina”) Genotype G:G NA NA 13 89.8 2 88.9 Heterozygous effect + 2.9 days + 1.7 days + 1.3 days (in number of days) P-value < 2.2e-16∗∗∗ 2.45e-05∗∗∗ 0.02790∗ KASP_9.958 Genotype C:C (as 55 91.9 48 91.1 21 86.6 “Garnet”) Genotype T:C (as 60 94.9 49 92.4 28 88.8 “Regina”) Genotype T:T NA NA 7 89.7 2 88.9 Heterozygous effect (in + 3.0 days + 1.3 days + 2.2 days number of days) P-value < 2.2e-16∗∗∗ 0.00101∗∗ 0.00015∗∗∗ Haplotype defined by A:A-C:C 55 91.9 45 91.1 19 87.0 KASP_9.936 and G:A-T:C 59 94.9 43 92.7 28 88.8 KASP_9.958 Heterozygous effect (in + 2.9 days + 1.7 days + 1.8 days number of days) P-value < 2.2e-16∗∗∗ 1.88e-05∗∗∗ 0.00266∗∗ significant differences were observed between those more reproducible over ten years, reducing the effect of individuals for KASP_9.936 (1.3 days) and KASP_9.958 the environment and therefore increasing heritability. (2.2 days). Significant differences were also observed when considering the haplotypes defined by both Linkage maps were improved with the RosBREED KASP_9.936 and KASP_9.958, heterozygous individuals cherry 6 + 9 K SNP array flowering 1.8 days later than homozygous ones (Table 4). The linkage maps developed using the RosBREED 6 + 9 K In conclusion, for both KASP markers, significant dif- SNP array were compared with the ones previously ferences were observed between heterozygous individu- obtained with the RosBREED 6 K SNP array [8]. Sizes als as “Regina” and homozygous individuals as “Garnet” of LGs in both maps were close and no marker rear- in all three panels. rangements were observed. The new maps contained many more markers (around four and three times more for “Regina” and “Garnet” parental maps, respectively) and average distance between markers and largest gaps Discussion were reduced. In general, LGs of “Regina” (R1 to R8) Flowering date evaluation and heritabilities were denser than LGs of “Garnet” (G1 to G8), with lower Within both populations, FD was not stable across years. average distance between markers. This result seems to This trait is highly dependent on temperatures and can be in accordance with the fact that “Regina” is highly show important inter-annual variations, resulting from heterozygous compared to other modern cultivars [25]. the variability of the chilling and heat accumulations Although using a larger number of markers allowed across years. It is known that advances in FD can be filling several gaps found in maps obtained with the explained by an early fulfillment of the HRs induced RosBREED 6 K SNP array [8], some large gaps remained by high temperatures in late winter [8]. In our exper- on LGs R6, G2 and G4, which may be caused by a lack iments, f lowering occurred earlier when temperatures of recombination events within these regions in the from October to January were low and temperatures in cultivars. February were significantly superior to the mean. This Recently, Calle et al. [26] presented parental linkage could be related to a better fulfilment of the CRs in late maps of cultivars “Vic” and “Cristobalina” using the autumn and winter, and of the HRs in late winter. Corre- RosBREED 6 + 9 K SNP array and a 161-individuals F1 lations between the three stages of FD and between years population. These maps respectively contained 910 SNPs were high in both populations. High heritability values covering 636.7 cM and 789 SNPs covering 666.0 cM. were found, in the same range as those estimated in However, only 324 and 310 SNP markers mapped at sweet cherry prior to this study [4,8,9]. Values were higher unique genetic positions in “Vic” and “Cristobalina” within Population #1 compared to Population #2. Data is parental maps, respectively [26]. In our study, the 10 | Horticulture Research, 2022, 9: uhac042 genotyping of 454 individuals with the RosBREED cherry “Regina” × “Lapins” (R × L) population [4,8] and Popula- 6 + 9 K array allowed to develop maps that contained tion #1 [8]. In both cultivars, the QTL mapped at the more markers mapping at unique positions (459 out of bottom of LG1. More recently, Calle et al. [9] detected 598 for the “Regina” parental map and 330 out of 446 for a major QTL (PVE up to 60.9%) at the bottom of the the “Garnet” parental map) and that were slightly shorter LG1 using several populations derived from the extra- than those developed by Calle et al. [26]. Therefore, the early blooming parental cultivar “Cristobalina” [12]. In maps we developed present a higher marker density. This our study, the QTL on LG1 of “Garnet” showed PVE values can be due to the number of individuals: we increased up to 17.1% in Population #1 (FF 2008), 8.8% in Population the probability of recombination events by using more #2 (FF 2018) and mapped at the end of the LG. Confidence individuals, leading to the development of genetic maps intervals of this QTL colocalize in “Garnet”, “Lapins” and with a reduced average distance between markers. “Cristobalina” and cover the chromosomal region known to carry DAM genes [9,18,20,28]. Genetic determinism of flowering date The QTL that explained the highest percentage of Few QTLs for FD were detected within Population #1 phenotypic variation within both populations was qP- with the single-year analyses, three on average, which FD4.1m on LG R4. The high significance of this locus could be related to the reduced size of the population. In was demonstrated in Dirlewanger et al. [4] and later comparison, more QTLs were detected with the single- in Castède et al. [8] using both R × L population and year analyses within Population #2, nine on average, and Population #1 in which it explained up to 31.8% and confidence intervals were smaller. The detection was 45.8% of the phenotypic variation of FF, respectively. highly improved in Population #1 when combining all Our study also demonstrated the high stability of qP- years together through the multi-year analysis. QTLs FD4.1m : this QTL alone was significant across all the were detected on almost all LGs (thirteen), with some growing seasons. In both single and multi-year analyses, of them accounting for a very small proportion of the the accuracy of the analyses for this major QTL was phenotypic variance, and confidence intervals of the significantly improved with Population #2. While the QTL major QTLs detected with the single-year analysis were region spanned 2 126 110 bp in Population #1 (multi-year reduced. Prior to our study, Castède et al. [8] performed analysis), it covered around 380 000 bp in Population #2, multi-year analysis on Population #1 with five years of consisting in a reduction of more than 1.7 Mb. This result measurements and detected QTLs on eleven LGs. Here, could be also explained by the denser genetic maps of the the addition of five years of FD measurements led to R × G Population #2 (obtained with the RosBREED cherry a significant reduction in the size of the QTLs confi- 6 + 9 K SNP array) as compared to those of Population #1 dence intervals and the detection of new minor QTLs (obtained with the RosBREED cherry 6 K SNP array) and (qP-FD5.1m and qP-FD8.1m ). With Population #2, confi- by KASP markers added in the region of the QTL. dence intervals were further reduced with the multi- Our study is the first one presenting QTL analyses year detection analysis. Working with a population con- using two different experimental designs obtained from taining more individuals increased the power and the a same cross, “Regina” × “Garnet”. It provides important accuracy of the detection. QTL analyses were very con- information from a methodological point of view, as sistent in both populations: the large majority of the we were able to compare QTL analyses performed on loci detected with Population #1 were significant with a small population of 117 individuals evaluated for FD Population #2 as well. The QTL qP-FD6.2m on LG G6 during ten years (Population #1) and a large population was an exception. This QTL was detected with Popula- of 454 individuals evaluated for FD during three years tion #1 across a few years and showed PVE values up (Population #2). When using a small population, single- to 14.6%, while it was not found in any of the analy- year QTL analyses allowed to detect few QTLs within ses performed with Population #2. Because it was not large confidence intervals. Multi-year analyses improved stable across years, the significance of this QTL could detections; however, as showed when comparing our be due to specific climatic conditions and/or genotype results with those presented by Castède et al. [8], a × environment interactions (in 2011 and 2016), which high number of years of phenotyping is required. In were not found in the three years of measurements in comparison, many more QTLs were detected in single- Population #2. year analyses when using a large population. This is Overall, the large number of QTLs detected confirmed particularly interesting for minor QTLs. Moreover, QTL that FD is a complex trait with a polygenic control. Most confidence intervals were reduced. Multi-year analyses of these QTLs were also found in peach and apricot [4]. further enhance the accuracy of the QTL detections, QTLs on LGs G1 (qP-FD1.2m ) and R4 (qP-FD4.1m ) were the even if only a few years of phenotyping (three years only loci with PVE values higher than 5% to be found in for Population #2) are available. In the present study, common in both populations. Both were firstly identified using a large population significantly improved the QTL in sweet cherry by Dirlewanger et al. [4] and have also analyses. However, two different SNP arrays were used to been reported in other Prunus species including peach, genotype Populations #1 and #2, therefore, genotyping is apricot, almond and sour cherry [10,13,16,27]. The QTL also of main importance. Our study showed that using on LG1 was detected in “Lapins” and “Garnet” using a a large number of individuals genotyped with a large Branchereau et al. | 11 amount of markers leads to more accurate QTLs. This as possible would allow breeders to optimize marker- strategy is particularly adapted to the study of agro- assisted selection (MAS) for different key agronomical nomical traits that can be quickly scored, such as FD. traits simultaneously. However, since the period covered by the f lowering of a sweet cherry population is in general less than four Candidate genes for flowering date within the weeks, phenotyping might become challenging if several QTL on LG4 thousands of progenies were to be scored. While several studies have confirmed that the QTL for FD on LG1 contains DAM genes [18–20,27,37], little is Fine mapping of qP-FD4.1m , a major flowering known about the QTL on LG4. In almond, the major date QTL on LG4 gene Late blooming (Lb) was identified in LG4 [38] but no The first objective of fine mapping is to narrow a given candidate genes co-localizing with this locus have been QTL region in order to reduce the number of candidate successfully found [39]. In sweet cherry, to date, the most genes. This strategy has already been used in peach for promising candidate genes are related to gibberellins and several traits, among them fruit acidity [29] (fine map- temperature sensing pathways [20]. They were identified ping of the D locus within a 100 kb region), maturity date by using the peach genome sequence annotation. [30] (fine mapping of the major LG4 locus in a 220 kb In our study, we used the “Regina” genome sequence interval), plant height [31] (fine mapping of the Tssd gene and identified 65 new candidate genes, among which in a 500 kb region), powdery mildew resistance [32] (fine 46 were expressed in our plant material. QTL analy- mapping of the Vrn3 gene in a 270 kb region) and skin ses with Population #2 allowed us to reduce the size fuzziness [33] (fine mapping of the G locus in a 481 kb of the QTL interval and none of the candidate genes interval). selected in Castède et al. [20] mapped within this refined In our study, we aimed to precisely map the sweet region. Two of them, EMF2 (EMBRYONIC FLOWER2) and cherry FD QTL qP-FD4.1m that we detected in a 380 kb- NUA (NUCLEAR PORE ANCHOR), were included in the new interval on LG4 with MultiQTL, between markers map and were found far from our QTL region (Fig. 2). KASP_9.780 and KASP_10.159. With the fine mapping, We could reduce our 46 candidate genes list down to a we concluded that it was located between KASP_9.890 set of twelve genes based on fine mapping. Several genes, and KASP_9.958, within a 68 kb-region. Developing differentially expressed in “Regina” and “Garnet”, could tightly linked KASP markers and phenotyping the whole be considered as promising, such as PavBOI-E3, PavSR45a Population #2 allowed us to improve the QTL localization. and PavSAUR71. In Arabidopsis, the RING domain E3 lig- However, we must remain cautious and take into consid- ase BOI represses flowering by repressing the expression eration several limits in our experiments. First of all, the of FLOWERING LOCUS T (FT) by two different ways: BOI phenotypic differences were small. Although “Regina” binds to CONSTANS (CO) to inhibit its targeting to FT; or and “Garnet” were well differentiated, late and early BOI targets FT via DELLA proteins [40]. Both mechanisms flowering respectively, hybrids with non-recombinant result in decreased expression of FT mRNA and inhibit parental genotypes differed by only three days on flowering. Based on our transcriptomic experiments, average. Small FD differences were observed between our PavBOI-E3 was more expressed in “Regina” than in recombinant genotypes as well. The whole population “Garnet” during endodormancy. This could be related tended to f lower rather early, closer to “Garnet” than to a stronger flowering inhibition in “Regina”. However, “Regina”. Unlike other important agronomic traits in after “Garnet” dormancy release, PavBOI-E3 expression peach from studies cited earlier, FD is not controlled importantly increased in “Garnet” and became higher by only one major gene or QTL. Therefore, phenotypic than in “Regina”, which is not anymore in accordance variations which are non-related to the QTL on LG4 may with the predicted function of the gene and its effect on occur in our recombinants. This could be the case for FD. PavSR45a was also differentially expressed between example for recombinant genotypes #1, #2, #5 and #7, “Regina” and “Garnet” especially after the endodormancy which did not allow to precise the QTL position. Finally, release period. This gene encodes the serine/arginine- we only had 41 individuals with recombinant genotypes, rich (SR) protein SR45a, a splicing factor (PavSR45a). In among which 33 with contrasted phenotypes could be Arabidopsis thaliana, the loss-of-function mutant sr45–1 used to fine-map the QTL. The mapping resolution exhibits pleiotropic phenotypes, among them a late depends on the number of recombinants and even with a flowering phenotype and elevated levels of FLOWERING large effect QTL, the higher the number of recombinants, LOCUS C (FLC), a major flowering repressor [41] within the better the mapping accuracy. the Brassicaceae family. It was demonstrated that Fine mapping of this FD QTL in sweet cherry was of SR45 influences the autonomous flowering pathway main interest, in particular because LG4 is considered in a FLC-dependent way in Arabidopsis [41,42]. SR45 to be a hot spot QTL LG in sweet cherry. Indeed, it is has also been related to the epigenetic regulation of known to carry major QTLs associated to fruit firmness FLOWERING WAGENINGEN (FWA), another f lowering- [34], maturity date [35] and rain-induced fruit cracking related gene [43]. SR45 protein was also reported to [36]. All these QTLs map at distinct but close positions on affect the alternative splicing of other SR genes [41] the same LG. Therefore, precising their location as much and to negatively regulate sugar signaling by repressing 12 | Horticulture Research, 2022, 9: uhac042 glucose-induced ABA accumulation [44]. Therefore, SR45 plants and are reported to participate in numerous is an important splicing factor regulating genes involved processes including plant growth, development, stress in growth, development and response to environmental responses and senescence [54]. In Rosaceae species peach, changes. In our material, expression levels of this gene apple and sweet cherry, NAC TFs have been reported increased during dormancy, reached a peak and then as candidate genes for a maturity date QTL located decreased after “Regina” dormancy release in both in a distinct region of LG4 [30,35,55]. In Arabidopsis, cultivars. PavSR45a was slightly more expressed in JUB1 TF represses GA3ox1 and DWF4 genes involved in “Regina”, especially during the late sampling dates of gibberellins and brassinosteroids biosynthesis pathways, both transcriptomic analyses. Moreover, the expression leading to reduced levels of these hormones and the drop was faster in “Garnet”. We could hypothesize that accumulation of DELLA proteins, restricting plant growth several f lowering-related genes are down-regulated by while promoting stress tolerance [56]. JUB1 delays SR45. Hence, the lower expression level observed in senescence, modulates cellular H2 O2 levels and also “Garnet” could lead to an earlier f lowering as compared enhances various abiotic stress tolerance responses to “Regina”. PavSAUR71 is also a promising CG. SAUR by targeting DREB2A [57]. Recently, ABA3, JUB1 and genes constitute the largest family of early auxin- COL3 were found to be under positive selection during responsive genes and play crucial roles in plant growth apricot domestication, most likely for selection on tree and development control [45]. In both “Regina” and phenology and environment adaptation [58]. “Garnet” cultivars, expression of this gene was null during endodormancy and started to increase after dormancy release. This is in accordance with increases of KASP markers usable in marker-assisted auxin levels during dormancy release reported in several selection studies [46]. PavSAUR71 expression levels increased For two KASP markers, KASP_9.936 and KASP_9.958, we more importantly in “Garnet” and could be related to found that the heterozygous accessions (as “Regina”) its early f lowering phenotype, compared to “Regina”. were significantly later flowering than homozygous ones Several genes encoded a G-type lectin S-receptor-like (as “Garnet”) using two validation panels and a F1 popu- serine threonine-kinase (GsSRK) that regulates both lation. Phenotypic differences between heterozygous and plant architecture and salt stress responses [47]. A homozygous individuals were much more important in gene encoding a kelch-repeat domain containing F-box the R × L population (three days) than in both panels protein (KFB) was also found within the fine mapping (between one and two days). It is known that the QTL interval. In Arabidopsis, several KFBs are known to be on LG4 is the major locus in this population as well involved in circadian clock and photoperiodic f lowering [4,8]. However, the germplasm collection and the panel of time regulation [48]. However, the expression profiles of cultivars exhibit a much larger genetic diversity. A large this gene were not conclusive. number of QTLs are likely to be involved in the control of Although they were not located within the fine map- FD and the QTL on LG4 may not be the major FD QTL in ping confidence interval, some genes with relevant func- this material. This could explain why the effect of this tions were worthy to consider. PAV04_REGINAg0203151 QTL was lower in these two panels. Nevertheless, this encodes the CONSTANS-LIKE 3 (COL3) transcription result can be useful in a MAS program. factor (TF) from the CCT (CO, CO-like and TOC1) family. In the diversity panels (germplasm collection and To date, genes from the CCT family are manly described cultivars), some individuals had an additional homozy- in cereal crops, where they are involved in the control gous genotype compared to the population. In the set of flowering time in response to the photoperiod and of 51 cultivars, those individuals (two in number, that is the circadian clock [49–51]. In trees, reductions in day 4% of the panel) presented a late flowering phenotype. length have been shown to induce growth cessation They were slightly later flowering than heterozygous and bud dormancy, however, the molecular mechanisms individuals for both KASPs. This was in accordance underlying the effects of photoperiod on growth and with what we expected from allele combinations: for developmental transitions remain quite unclear [52]. instance for KASP_9.958, homozygous C:C are early FD, PAV04_REGINAg0203191 codes for the enzyme ABA3, heterozygous T:C are late flowering, and homozygous T:T the molybdenum cofactor sulfurase involved in the are further late flowering. Different results were found synthesis of the sulfureted form of the molybdenum in the germplasm collection. Firstly, more individuals cofactor which is required for the activity of molybde- presented the new homozygous genotype (6% for num enzymes such as aldehyde oxidase (AO) [53]. AO KASP_9.958 and 12% for KASP_9.936), confirming that catalyzes the final step of the biosynthesis of abscisic a larger genetic diversity is present in the collection. acid (ABA). Therefore, ABA3 plays an essential role Interestingly, these individuals were flowering earlier in the biosynthesis of ABA, a major plant hormone than those having the other homozygous genotype promoting seed and bud dormancy [46,53]. Finally, (as “Garnet”). One explanation could be that these PAV04_REGINAg0203201 encodes the JUNGBRUNNEN individuals have the early allelic combination at other FD 1 (JUB1) TF from the NAC (NAM, ATAF, CUC) family. QTLs. Moreover, in these genetic backgrounds, the QTL NAC TFs constitute one of the largest TF families in on LG4 might not be the one explaining the largest PVE. Branchereau et al. | 13 FD is a quantitative trait controlled by many genomic composed of 1386 hybrids. The cross was made in 2010 regions. The major QTL on LG4 is one of them, but using potted trees (five trees of each parental cultivar) other regions such as LG1 play an important role too and bumblebees in confined tunnels. Among the 1386 [9]. Therefore, markers should be designed in these other hybrids created, 793 had “Regina” (R × G hybrids) while important genomic regions. Recently, Calle et al. [28] 593 had “Garnet” (G × R hybrids) as maternal parent, developed two DNA-based markers in the QTL on LG1, respectively. Population #2 was planted in 2012 in the within the DAM genes regions. Developed from the extra- same experimental site as Population #1. Trees were early cultivar “Cristobalina”, these markers could be use- planted every two meters in rows separated by five ful for selection for early f lowering and low CRs in sweet meters. cherry. Combined together, these new markers should For the KASP marker validation, a population derived allow the deployment of a complete MAS strategy in from a cross using “Regina” and “Lapins” (R × L) and two sweet cherry for FD. sets of accessions were used (Table S10). The R × L pop- ulation is composed of 115 individuals planted on their own roots in the UEA of INRAE, in Toulenne. The first set Conclusion of accessions is a subset of 104 accessions from the sweet We report in this study the first fine mapping performed cherry core collection defined from the INRAE sweet in sweet cherry, for the major FD QTL on LG4. The QTL cherry germplasm collection, maintained by the INRAE’s on LG4 was highly stable in our plant material, which has Prunus Genetic Resources Center in Bourran (120 km relatively high CRs and is rather late blooming. Therefore, south-east from Bordeaux, France) [59]. This panel is this region is of main interest in sweet cherry breeding already well characterized and we carefully selected indi- programs as well as the QTL on LG1 associated to low viduals representing a large genetic diversity and cover- chilling cultivars to create new cultivars well adapted to ing a large variability for FD [59]. The second one is a their growing area. Our results constitute the first step set of 51 cultivars, including modern cultivars, planted for the development of a set of markers within the LG4 in Toulenne, in the same area as the three other popula- QTL that could be used in MAS for FD in sweet cherry. tions. We reduced the 380-kb region obtained with the new QTL For the RNA-seq analyses, flower bud samples were analyses using a large population to an 68-kb region con- collected during two seasons from “Garnet” and “Regina” taining only twelve candidate genes. The most likely can- trees grown at the INRAE UEA in Toulenne (Decem- didate genes, with interesting expression patterns, were ber 2009–March 2010) and Bourran (July 2015 – March related to splicing (SR45a) and auxin-response (SAUR71). 2016). As previously described in Vimont et al. [24], a Further analyses based on transformation experiments mix of randomly chosen flower buds (equivalent to a on model species could be performed to validate the 2 mL volume) were harvested from branches of two robust candidate genes we identified and might give or three different trees, corresponding to the biological new insights into the control of FD in Prunus species. replicates. Upon harvesting, buds were flash frozen in Moreover, our study provides relevant information from liquid nitrogen and stored at −80◦ C prior to performing a methodological point of view by using a same cross RNA-seq. in two different experimental designs to compare QTL analyses. Flowering date phenotyping Three flowering stages were scored: beginning of flow- Materials and methods ering (BF), when approximately 10% of the floral buds Plant material reached full bloom; full flowering (FF), when 75% of the Two F1 sweet cherry populations derived from crosses floral buds reached full bloom; end of flowering (EF), using “Regina” and “Garnet” cultivars were analyzed when more than 50% of the flowers were wilting. Trees (Table 1). “Regina” is a late blooming German cultivar, were observed from three to four times a week during whereas “Garnet” is an early blooming cultivar from the the season to score the different flowering stages in USA. The first population, called hereafter “Population Julian days, JDs. Within Population #1, FD was evaluated #1”, is composed of 117 individuals obtained from the during ten years from 2008 to 2017. The first five years cross “Regina” × “Garnet”. Trees were planted in 2003 on of evaluation have been utilized in Castède et al. [8]. In their own roots (not grafted, therefore planted without the present study, five additional years from 2013 to 2017 replication) in the Tree Experimental Unit (UEA) of the were used to refine the QTL analyses. Within Population French National Research Institute for Agriculture, Food #2, BF and FF were evaluated during three years, in 2018, and the Environment (INRAE)-Bordeaux research center, 2019 and 2021 while EF was evaluated in 2018 and 2019 at Toulenne (50 km south-east from Bordeaux, France). but not in 2021 because of frost events in March–April Trees were planted every 2.5 meters in rows separated by preventing a reliable evaluation (Table 1). six meters, orchards were not irrigated. This population For KASP validation, FD (BF stage) was scored during has been firstly used and presented in Castède et al. ten years in the R × L population, from 2006 to 2016 [8]. The second population, called “Population #2”, is (Table S10). The two sets of accessions were evaluated for 14 | Horticulture Research, 2022, 9: uhac042 BF during six years, from 2014 to 2019, each accession analyses and marker filtering were done using GenomeS- being evaluated at least three years (Table S10). tudio software (v2.0, Illumina) as described in Klagges Daily temperatures were collected in the orchard et al. [62]. Monomorphic markers, individuals with more located at Toulenne using an automatic data-logger than 5% missing data and markers with more than 10% (Ebro® ; Ebro Electronic, Ingolstadt, Germany) in order missing data were discarded to construct linkage maps of to characterize environmentally the years of evaluation. each parent “Regina” and “Garnet”, following the pseudo test-cross methodology used for heterozygous species Measurements of bud break and estimation of [63]. JoinMap® software [64] (version 4.1) was used to the dormancy release date perform linkage analysis using SNPs which were het- Measurements for the dormancy stages were performed erozygous in only one of the two parents (classes coded on randomly chosen branches cut every two weeks from and < nn x np>). The independence LOD test November 16th 2015 to April 4th 2016 for “Garnet” and (threshold = 15.0) and the regression mapping function “Regina”. Branches were incubated in water pots placed were used for markers grouping and maps construc- in a growth chamber (25◦ C, 16 h light/8 h dark, 60–70% tion, respectively. SNPs with identical segregation in the humidity). The water was replaced every 3–4 days. After population were included in the maps (function “Assign ten days under these forcing conditions, the percentage identical loci to their groups”). of bud break, i.e. f lower buds at BBCH stage 53 [60] was QTL analyses for flowering date recorded. The date of dormancy release was estimated when at least 50% of the f lower buds were at the BBCH QTL analyses were based on linkage maps already avail- stage 53 or higher after ten days under forcing conditions. able for Population #1 [8] and on the new maps for Population #2. For both populations, QTL analyses were performed for the three FD stages using the Multiple Statistical analysis Interval Mapping (MIM) method implemented in Mul- Distribution, mean, minimum and maximum values of tiQTL V2.6 software (http://www.multiqtl.com). Detec- BF, FF and EF were estimated for each year. Addition- tions were carried out separately for “Regina” and “Gar- ally, Spearman correlation coefficients between the three net” parental maps by using the “single QTL model” flowering stages and between years were calculated. (one-QTL per linkage group). We tested the “two-linked Analyses were performed using “ggplot2” and “psych” QTLs model” but no consistent detections were observed R packages. Broad-sense heritability (H2 ), to measure (results not shown). Both single-year (or year-by-year) the between-year stability of the f lowering traits, was and multi-year models were utilized (multi-environment also estimated from the analysis of variance based on model available in MultiQTL). When performing multi- the following mixed model, as previously described in year analysis, a single position (in cM) and a single LOD Dirlewanger et al. [4]: value are given while values of percentage of variation Yij = μ + gi + yj + eij . explained (PVE) are estimated for each year. For ease of reading, the mean PVE value across years is presented. where Yij is the phenotypic value of the ith individual in This study gives an update of the QTL analyses presented the jth year, μ is the mean value of the trait, gi is the in Castède et al. [8] by using five additional years of FD random genotypic effect of individual i, yj is the fixed phenotyping to perform a multi-year analysis with the effect of year j and eij is the residual of the model (i.e. 10 years available in total, from 2008 to 2017. Graphical genotype × year interaction). This linear mixed-effects representation of QTLs on linkage maps was generated model was fitted in R using the lme4 package [61]. with MapChart software v2.3 [65]. Only QTLs for FF were Heritability was then estimated using the following presented in this paper. In accordance with to the GDR equation: nomenclature (see www.rosaceae.org for more details), 2 σg H2 =  2  they were called “qP-FD”. 2 + σe σg n Fine mapping of the QTL on “Regina” LG4 with where σg2 is the genetic variance, σe2 the residual variance KASP markers (environmental variance) and n is the number of years. SNPs located within the confidence interval of the FD Heritabilities were calculated for both populations. QTL on LG4 were selected from different sources: i) SNPs within the RosBREED cherry 6 + 9 K Illumina Infinium® Genotyping and linkage mapping SNP array, based on their position on the sweet cherry Genomic DNA from parental lines and from hybrids from physical map; ii) SNPs from GBS analyses described in Population #2 was extracted from young leaves. A subset a previous study [66]; and iii) SNPs identified through of 454 R × G hybrids (randomly chosen) from Population the mapping of available “Regina” RNA-sequencing #2 was genotyped with the RosBREED cherry 6 + 9 K data on the “Regina” genome sequence (Table S4). In Illumina Infinium® SNP array [23]. The strategy consist- this last case, SNPs were identified using IGV software ing in the genotyping of a subset of Population #2 was (http://software.broadinstitute.org/software/igv/), and followed to reduce the cost of the analyses. SNP genotype heterozygous markers with high RNA-seq coverage were Branchereau et al. | 15 selected. Well-distributed SNPs were then selected to be in 2009/2010 and 2015/2016 on cultivars “Regina” and transformed into Kompetitive Allele Specific PCR (KASP) “Garnet”, in order to obtain the expression profiles of the markers based on the dual Fluorescence Resonance candidate genes. Energy Transfer (FRET) method, in which the sample DNA During the 2009/2010 sampling season, three dates is amplified with allele specific primers conjugated to were chosen for RNA sequencing: 3 December 2009, fluorometric dyes HEX and FAM at their 5 end. For each 1 February 2010 and 16 March 2010, associated with KASP, three primers were developed by the BioGEVES different stages of dormancy, respectively endodor- laboratory (Beaucouzé, France) and the reactions were mancy, endodormancy release and ecodormancy. For performed as described in Bernard et al [67]. both cultivars, three biological replicates were used at Fine mapping of the QTL region on LG R4 was carried the three dates. Total RNA was extracted from 50 mg out into two steps. Firstly, nine KASPs were used to of frozen ground flower buds and sequenced on an genotype the complete Population #2 (1386 individu- Illumina® HiSeq 2000 (single read) by GATC Biotech als): KASP_9.269, KASP_9.271, KASP_9.456, KASP_9.727, (Mulhouse, France) in 2011. For the 2015/2016 samples, KASP_9.780, KASP_9.890, KASP_9.935, KASP_10.090 and total RNA was extracted and sequenced as described in KASP_10.159 (named accordingly to their physical posi- Vimont et al. [24]. Eleven dates spanning from July 2015 tion on sweet cherry LG4 in kb). Recombinant individuals to the end of March 2016 were used, including the three detected in this region were then genotyped with eight mentioned dormancy stages covered with the 2009/2010 new KASP markers, selected on IGV software and located sampling. Sequencing data are available online (Bio- within genes between KASP_9.780 and KASP_10.159, Project PRJNA756935 and Gene Expression Omnibus named KASP_9.781 (gene PAV04_REGINAg0203151), GSE130426). The quality of raw reads was assessed KASP_9.801 (gene PAV04_REGINAg0203181), KASP_9.814 using FastQC (www.bioinformatics.babraham.ac.uk/pro (gene PAV04_REGINAg0203191), KASP_9.916 (gene PAV04_ jects/fastqc/) and possible adaptor contaminations and REGINAg0203361), KASP_9.933 (gene PAV04_REGINAg02 low quality trailing sequences were removed using 03391), KASP_9.936 (gene PAV04_REGINAg0203391), Trimmomatic [70]. Raw reads sequences were mapped KASP_9.958 (gene PAV04_REGINAg0203421) and KASP_ on the sweet cherry “Regina” reference genome [68] (v1.0) 9.970 (in the gene PAV04_REGINAg0203451). Least square using STAR as previously described in Vimont et al. [71]. means (lsmeans) of the three years of data (2018, 2019 Raw counts and transcripts per million reads (TPM) for and 2021) for FD were calculated for the parents, the each transcript were calculated using HTSeq [72,73]. non-recombinant genotypes and the groups of same recombinant genotypes. With this fine mapping of the KASPs validation QTL, our objective was to reduce the confidence interval Four KASP markers from the set of the seventeen of its position and, further, to reduce the number of markers used for the fine mapping, were tested for putative candidate genes. validation on different genetic backgrounds. We selected markers among those included within the QTL interval In silico candidate genes identification established by our fine mapping approach, spanning Chromosomal region for in silico CG analysis was selected from KASP_9.890 to KASP_9.958: KASP_9.916, KASP_9.936 based on the QTL on LG4 detected with MultiQTL within and KASP_9.958. Moreover, although it was outside Population #2 (QTL analyses using 454 R × G hybrids). the interval, we also selected KASP_9.814 as several Predicted cherry gene models and corresponding protein recombination events occurred between this marker sequences of this genomic region were retrieved from and KASP_9.890 (eight individuals out of 41 were “Regina” sweet cherry genome repository [68] (https:// recombinant between these markers). Recombination doi.org/10.15454/KEW474). Data mining on the gene events occurred between KASP_9.890 and KASP_9.916 for ontology terms associated with candidate genes was nine individuals, therefore, we also selected KASP_9.916. done using Blast2GO Version 1.4.4 [69]. Predicted cherry No recombination occurred between KASP_9.916 and peptides were used for similarity search in a non- KASP_9.933, so we did not select KASP_9.933. A recom- redundant genebank protein database with blastp bination occurred between KASP_9.933 and KASP_9.935. algorithm with a minimum e-value <10−6 before gene However, as KASP_9.935 and KASP_9.936 were very close ontology mapping and annotation. When no functional and no recombination event occurred between them, we annotation from Blast2GO was available, an annotation decided to select KASP_9.936. Finally, one recombination from the Arabidopsis database TAIR (The Arabidopsis event occurred between KASP_9.936 and KASP_9.958, Information Resource) was added. so we selected KASP_9.958. In the end, KASP_9.814, KASP_9.916, KASP_9.936 and KASP_9.958 were tested for RNA-seq data and analyses validation on the R × L population and on the two sets of Transcriptomic data was used to look for differentially accessions (Table S10). expressed genes in cultivars “Regina” and “Garnet” which In order to determine if FD was different between could potentially explain the FD differences we observed. heterozygous and homozygous genotypes, analyses of We used the results of two experiments of whole variances (ANOVAs) were performed in R software. The transcriptome analyses (RNA-sequencing) performed following linear model was used: 16 | Horticulture Research, 2022, 9: uhac042 Yij = μ + gi + yj + eij where Yij is the phenotypic value of Supplementary data the ith individual in the jth year, μ is the mean value of Supplementary data is available at Horticulture Research the trait, gi is the random genotypic effect of individual i, online. yj is the fixed effect of year j and eij is the residual of the model. References Acknowledgements 1. Quero-García J, Schuster M, López-Ortega G, Charlot G. Sweet cherry varieties and improvement. In: Quero-García J, Iezzoni A, This work was supported by the “Région Nouvelle- Pulawska J, Lang G, eds. Cherries: Botany, Production and Uses. CABI Aquitaine” in the project “CerGEn” reference 2018- (UK), 2017,60–94. 1R20203, numbers 00019006 and 00019007. INRAE BAP 2. Radičević S, Cerović R, Nikolić D, Ðord̄ević M. The effect of division and the “Région Nouvelle-Aquitaine” co-funded genotype and temperature on pollen tube growth and fertil- the PhD scholarship of C.B and funded the Postdoctoral ization in sweet cherry (Prunus avium L.). Euphytica. 2016;209: fellowship of A.D. The “Groupement d’Intérêt Scien- 121–36. tifique Fruits”, GIS Fruits (https://www.gis-fruits.org/), 3. Wenden B, Campoy JA, Jensen M, López-Ortega G. Climatic limit- funded the Master thesis scholarship of N.H.Z.E. ing factors: temperature. In: Quero-García J, Iezzoni A, Pulawska We thank Hélène Christmann and Jacques Joly for J, Lang G, eds. Cherries: Botany, Production and Uses. CABI (UK), 2017, their contribution to the phenotyping of the f lowering 166–88. 4. Dirlewanger E, Quero-García J, Le Dantec et al. Comparison of the date of the accessions of the INRAE’s Prunus Genetic genetic determinism of two key phenological traits, flowering Resources Center in Bourran and of the cultivars panel in and maturity dates, in three Prunus species: peach, apricot and Toulenne, respectively. We thank INRAE’s Experimental sweet cherry. Heredity (Edind). 2012;109:280–92. Unit UEA for the management of the populations and 5. Lang GA, Early JD, Martin GC, Darnell RL. Endo-, Para-, and cultivars in Toulenne and for preserving and managing ecodormancy: physiological terminology and classification for the collections of the INRAE’s Prunus Genetic Resources dormancy research. HortScience. 1987;22:371–7. Center in Bourran. 6. Luedeling E. Climate change impacts on winter chill for tem- We also acknowledge the BioGEVES laboratory for DNA perate fruit and nut production: a review. Sci Hortic. 2012;144: extraction and KASP marker development. 218–29. 7. Couvillon GA, Erez A. Influence of prolonged exposure to chilling temperatures on bud break and heat requirement for bloom of Author contributions several fruit species. J Am Soc Hortic Sci. 1985;110:47–50. 8. Castède S, Campoy JA, García JQ et al. Genetic determinism of J.Q-G. and E.D. designed the experiments and provided phenological traits highly affected by climate change in Prunus financial support; D.A. and J.P. carried out phenotyping; avium: flowering date dissected into chilling and heat require- C.B. analyzed the data; N.H.Z.E., L.L., B.W., A.D., M.F., L.L.D ments. New Phytol. 2014;202:703–15. and T.B. made contributions to data analysis; C.B., J.Q-G. 9. Calle A, Cai L, Iezzoni A, Wünsch A. Genetic dissection of bloom and E.D. wrote the manuscript. time in low chilling sweet cherry (Prunus avium L.) using a All authors reviewed and approved the final manuscript. multi-family QTL approach. Front Plant Sci. 2020;10:1647. 10. Sánchez-Pérez R, Del Cueto J, Dicenta F, Martínez-Gómez P. Recent advancements to study flowering time in almond and other Prunus species. Front Plant Sci. 2014;5:334. Data availability 11. Dirlewanger E, Graziano E, Joobeur T et al. Comparative mapping Phenotypic data are available in the excel file “Raw data and marker-assisted selection in Rosaceae fruit crops. Proc Natl - Phenotyping data”. Acad Sci U S A. 2004;101:9891–6. Assembly and annotation of the “Regina” sweet cherry 12. Alburquerque N, García-Montiel F, Carrillo A, Burgos L. Chilling genome is available at https://data.inrae.fr/dataset. and heat requirements of sweet cherry cultivars and the rela- xhtml?persistentId=doi:10.15454/KEW474 tionship between altitude and the probability of satisfying the chill requirements. Environ Exp Bot. 2008;64:162–70. RNA-seq data from 2015-2016 have been deposited in 13. Sánchez-Pérez R, Dicenta F, Martínez-Gómez P. Inheritance of the NCBI Gene Expression Omnibus under the accession chilling and heat requirements for flowering in almond and QTL code GSE130426. analysis. Tree Genet Genomes. 2012;8:379–89. RNA-seq data from 2009–2010 have been deposited 14. Campoy JA, Ruiz D, Allderman L et al. The fulfilment of chilling in the NCBI Short Read Archive under the accession requirements and the adaptation of apricot (Prunus armeniaca code PRJNA756935 (https://www.ncbi.nlm.nih.gov/sra/ L.) in warm winter climates: an approach in Murcia (Spain) and PRJNA756935). the Western cape (South Africa). Eur J Agron. 2012;37:43–55. 15. Okie WR, Blackburn B. Increasing chilling reduces heat require- ment for floral Budbreak in peach. HortScience. 2011;46: Conflict of interest 245–52. 16. Cai L, Stegmeir T, Sebolt A et al. Identification of bloom date The authors declared that they have no competing inter- QTLs and haplotype analysis in tetraploid sour cherry (Prunus ests. cerasus). Tree Genet Genomes. 2018;14:22. Branchereau et al. | 17 17. Tavaud M, Zanetto A, David JL et al. Genetic relationships 35. Calle A, Wünsch A. Multiple-population QTL mapping of between diploid and allotetraploid cherry species (Prunus maturity and fruit-quality traits reveals LG4 region as a avium, Prunus × gondouinii and Prunus cerasus). Heredity. breeding target in sweet cherry (Prunus avium L.). Hortic Res. 2004;93:631–8. 2020;7:1–13. 18. Bielenberg DG, Wang Y(E, Li Z et al. Sequencing and annotation 36. Quero-García J, Letourmy P, Campoy JA et al. Multi-year analyses of the evergrowing locus in peach [Prunus persica (L.) Batsch] on three populations reveal the first stable QTLs for tolerance reveals a cluster of six MADS-box transcription factors as can- to rain-induced fruit cracking in sweet cherry (Prunus avium L.). didate genes for regulation of terminal bud formation. Tree Genet Hortic Res. 2021;8:136. Genomes. 2008;4:495–507. 37. Bielenberg DG, Rauh B, Fan S et al. Genotyping by sequencing 19. Zhebentyayeva TN, Fan S, Chandra A et al. Dissection of chilling for SNP-based linkage map construction and QTL analysis of requirement and bloom date QTLs in peach using a whole chilling requirement and bloom date in peach [Prunus persica (L.) genome sequencing of sibling trees from an F2 mapping pop- Batsch]. PLoS One. 2015;10:e0139406. ulation. Tree Genet Genomes. 2014;10:35–51. 38. Socias R, Felipe AJ, Aparisi JG. A major gene for flowering time 20. Castède S, Campoy JA, Le Dantec et al. Mapping of candidate in almond. Plant Breed. 1999;118:443–8. genes involved in bud dormancy and flowering time in sweet 39. Silva C, Garcia-Mas J, Sánchez AM et al. Looking into flowering cherry (Prunus avium). PLoS One. 2015;10:e0143250. time in almond (Prunus dulcis (mill) D. A. Webb): the candidate 21. Verde I, Abbott AG, Scalabrin S et al. The high-quality draft gene approach. Theor Appl Genet. 2005;110:959–68. genome of peach (Prunus persica) identifies unique patterns 40. Nguyen KT, Park J, Park E et al. The Arabidopsis RING domain of genetic diversity, domestication and genome evolution. Nat protein BOI inhibits flowering via CO-dependent and CO- Genet. 2013;45:487–94. independent mechanisms. Mol Plant. 2015;8:1725–36. 22. Peace C, Bassil N, Main D et al. Development and evaluation 41. Ali GS, Palusa SG, Golovkin M et al. Regulation of plant of a genome-wide 6K SNP Array for diploid sweet cherry and developmental processes by a novel splicing factor. PLoS One. tetraploid sour cherry. PLoS One. 2012;7:e48305. 2007;2:e471. 23. Vanderzande S, Zheng P, Cai L et al. The cherry 6+9K SNP array: 42. Chen SL, Rooney TJ, Hu AR et al. Quantitative proteomics a cost-effective improvement to the cherry 6K SNP array for reveals a role for SERINE/ARGININE-rich 45 in regulating RNA genetic studies. Sci Rep. 2020;10:7613. metabolism and modulating transcriptional suppression via 24. Vimont N, Fouché M, Campoy JA et al. From bud formation the ASAP complex in Arabidopsis thaliana. Front Plant Sci. to flowering: transcriptomic state defines the cherry develop- 2019;10:1116. mental phases of sweet cherry bud dormancy. BMC Genomics. 43. Ausin I, Greenberg MVC, Li CF, Jacobsen SE. The splicing factor 2019;20:974. SR45 affects the RNA-directed DNA methylation pathway in 25. Pinosio S, Marroni F, Zuccolo A et al. A draft genome of sweet Arabidopsis. Epigenetics. 2012;7:29. cherry (Prunus avium L.) reveals genome-wide and local effects 44. Carvalho RF, Szakonyi D, Simpson CG et al. The Arabidopsis of domestication. Plant J. 2020;103:1420–32. SR45 splicing factor, a negative regulator of sugar signaling, 26. Calle A, Serradilla MJ, Wünsch A. QTL mapping of phenolic modulates SNF1-related protein kinase 1 stability. Plant Cell. compounds and fruit colour in sweet cherry using a 6+9K SNP 2016;28:1910–25. array genetic map. Sci Hortic. 2021;280:109900. 45. Stortenbeker N, Bemer M. The SAUR gene family: the plant’s 27. Fan S, Bielenberg DG, Zhebentyayeva TN et al. Mapping quan- toolbox for adaptation of growth and development. J Exp Bot. titative trait loci associated with chilling requirement, heat 2019;70:17–27. requirement and bloom date in peach (Prunus persica). New 46. Liu J, Sherif SM. Hormonal orchestration of bud Phytol. 2010;185:917–30. dormancy cycle in deciduous Woody perennials. Front Plant 28. Calle A, Grimplet J, Le Dantec L, Wünsch A. Identification Sci. 2019;10:1136. and characterization of DAMs mutations associated with Early 47. Sun M, Qian X, Chen C et al. Ectopic expression of GsSRK in blooming in sweet cherry, and validation of DNA-based markers Medicago sativa reveals its involvement in plant architecture for selection. Front Plant Sci. 2021;12:621491. and salt stress responses. Front Plant Sci. 2018;9:226. 29. Boudehri K, Bendahmane A, Cardinet G et al. Phenotypic and fine 48. Naeem ul Hassan M, Zainal Z, Ismail I. Plant kelch containing genetic characterization of the D locus controlling fruit acidity F-box proteins: structure, evolution and functions. RSC Adv. in peach. BMC Plant Biol. 2009;9:59. 2015;5:42808–14. 30. Pirona R, Eduardo I, Pacheco I et al. Fine mapping and identifica- 49. Liu H, Zhou X, Li Q et al. CCT domain-containing genes in cereal tion of a candidate gene for a major locus controlling maturity crops: flowering time and beyond. Theor Appl Genet. 2020;133: date in peach. BMC Plant Biol. 2013;13:166. 1385–96. 31. Lu Z, Niu L, Chagné D et al. Fine mapping of the temperature- 50. Datta S, Hettiarachchi GHCM, Deng X-W, Holm M. Arabidopsis sensitive semi-dwarf (Tssd) locus regulating the internode CONSTANS-LIKE3 is a positive regulator of red light signaling length in peach (Prunus persica). Mol Breed. 2016;36:20. and root growth. Plant Cell. 2006;18:70–84. 32. Marimon N, Luque J, Arús P, Eduardo I. Fine mapping and 51. Tripathi P, Carvallo M, Hamilton EE et al. Arabidopsis B-BOX32 identification of candidate genes for the peach powdery mildew interacts with CONSTANS-LIKE3 to regulate flowering. PNAS. resistance gene Vr3. Hortic Res. 2020;7:1–9. 2017;114:172–7. 33. Lu Z, Pan L, Wei B et al. Fine mapping of the gene controlling 52. Singh RK, Svystun T, AlDahmash B et al. Photoperiod- and the fruit skin hairiness of Prunus persica and its uses for MAS in temperature-mediated control of phenology in trees – a molec- progenies. Plants (Basel). 2021;10:1433. ular perspective. New Phytol. 2017;213:511–24. 34. Cai L, Quero-García J, Barreneche T et al. A fruit firmness QTL 53. Bittner F, Oreb M, Mendel RR. ABA3 is a molybdenum cofac- identified on linkage group 4 in sweet cherry (Prunus avium L.) tor sulfurase required for activation of aldehyde oxidase and is associated with domesticated and bred germplasm. Sci Rep. xanthine dehydrogenase in Arabidopsis thaliana. J Biol Chem. 2019;9:5008. 2001;276:40381–4. 18 | Horticulture Research, 2022, 9: uhac042 54. Pereira-Santana A, Alcaraz LD, Castaño E et al. Comparative mapping strategy and RAPD markers. Genetics. 1994;137: genomics of NAC transcriptional factors in angiosperms: impli- 1121–37. cations for the adaptation and diversification of flowering 64. Van Ooijen JW. JoinMap® 4, software for the calculation of genetic plants. PLoS One. 2015;10:e0141866. linkage maps in experimental populations. Wageningen: Kyazma BV, 55. Migicovsky Z, Gardner KM, Money D et al. Genome to phenome 2006. mapping in apple using historical data. Plant Genome. 2016;9: 65. Voorrips RE. MapChart: software for the graphical presentation 1–15. of linkage maps and QTLs. J Hered. 2002;93:77–8. 56. Shahnejat-Bushehri S, Tarkowska D, Sakuraba Y, Balazadeh 66. Dirlewanger E, Le-Dantec L, Fouché M et al. New sweet cherry S. Arabidopsis NAC transcription factor JUB1 regulates GA/BR genomic tools and their use in marker-assisted breeding. Acta metabolism and signalling. Nat Plants. 2016;2:1–9. Hortic. 2020;533–42. 57. Wu A, Allu AD, Garapati P et al. JUNGBRUNNEN1, a reactive 67. Bernard A, Marrano A, Donkpegan A et al. Association and oxygen species-responsive NAC transcription factor, regulates linkage mapping to unravel genetic architecture of phenological longevity in Arabidopsis. Plant Cell. 2012;24:482–506. traits and lateral bearing in Persian walnut (Juglans regia L.). 58. Groppi A, Liu S, Cornille A et al. Population genomics of apricots BMC Genomics. 2020;21:203. unravels domestication history and adaptive events. Nat Com- 68. Le Dantec L, Girollet N, Jérôme G et al. Assembly and annotation mun. 2021;12:3956. of ‘Regina’ sweet cherry genome. 2020. 59. Campoy JA, Lerigoleur-Balsemin E, Christmann H et al. Genetic 69. Conesa A, Götz S, García-Gómez JM et al. Blast2GO: a universal diversity, linkage disequilibrium, population structure and con- tool for annotation, visualization and analysis in functional struction of a core collection of Prunus avium L. landraces and genomics research. Bioinformatics. 2005;21:3674–6. bred cultivars. BMC Plant Biol. 2016;16:49. 70. Bolger AM, Lohse M, Usadel B. Trimmomatic: a flexible trimmer 60. Fadón E, Herrero M, Rodrigo J. Flower development in sweet for Illumina sequence data. Bioinformatics. 2014;30:2114–20. cherry framed in the BBCH scale. Sci Hortic. 2015;192:141–7. 71. Vimont N, Schwarzenberg A, Domijan M et al. Fine tuning of 61. Bates D, Mächler M, Bolker B, Walker S. Fitting linear mixed- hormonal signaling is linked to dormancy status in sweet cherry effects models using lme4. J Stat Softw. 2015;67:1–48. flower buds. Tree Physiol. 2021;41:544–61. 62. Klagges C, Campoy JA, Quero-García J et al. Construction and 72. Anders S, Pyl PT, Huber W. HTSeq–a python framework to work comparative analyses of highly dense linkage maps of two sweet with high-throughput sequencing data. Bioinformatics. 2015;31: cherry intra-specific progenies of commercial cultivars. PLoS 166–9. One. 2013;8:e54743. 73. Wagner GP, Kin K, Lynch VJ. Measurement of mRNA abundance 63. Grattapaglia D, Sederoff R. Genetic linkage maps of Eucalyptus using RNA-seq data: RPKM measure is inconsistent among sam- grandis and Eucalyptus urophylla using a pseudo-testcross: ples. Theory Biosci. 2012;131:281–5.